To add: normally(oh, what a pun!) floating point values are stored "normalized", which means that 10.1234567 is actually 1.01234567e1 (well, actually the number is normalized in binary format, not decimal and the first 1 is not stored, as it is always there, except for 0,... ), so the 6th digit after the decimal point is 5, not 6.
On the other hand, 0.011234567 normalized is 1.1234567e-2, so you do "gain precision".